What is going on:
Maurizio has been working on shared-tabbles: document management p2p technology (more info here and here).
Basically, instead of loading only one database, Tabbles would load several databases, one for each user, have them all in memory at the same time, and whenever a query was done (e.g.: showing the content of a tabble or a combination), Tabbles would query all the databases and merge the output of the queries.
The great question lying behind this approach was: will this merge badly slow down Tabbles? And based on our first test we believe that the answer is a BIG FAT „NO“!